Hydrologische impact op de koolstofdynamiek van grote tropische riviersystemen

Rivers form an important connection between the three major pools of the fast cycling domain of the global carbon (C) cycle, i.e., the terrestrial, atmospheric and oceanic C pools. In initial representations of the global C cycle, rivers have been represented as inactive pipes which simply transfer C from the terrestrial pool to the oceans. However, it has now been recognized that a significant amount of C processing takes place in rivers systems, which can influence the distribution of C in the global system. Changes in the functioning of a river system due to climate change or anthropogenic impact such as land use changes or water retention, can either enhance or suppress C fluxes from and by the river systems. To gain a better understanding of the role of rivers, measurements of C concentrations and fluxes have been carried out for many river systems. Tropical river systems are usually underrepresented in global datasets. Nevertheless, they are important in view of their large contribution to global riverine discharge and their location in very productive ecosystems. In addition, those river systems are often characterized by contrasting dry and wet seasons which can alter the functioning of the system completely. The goal of this study was to increase our understanding of the C processing and fluxes that occur in tropical rivers with particular attention for the wet season. Therefore, sampling campaigns on the lower Tana River, Kenya were conducted during three different wet seasons (Oct-Nov 2012, May-June 2013, April-May 2014). The wet season of 2013 was characterized by extensive flooding between the two sampling sites, while the discharge peaks in 2012 and 2014 did not cause significant flooding. Daily water samples (n2012=37, n2013=52, n2014=45) were taken at Garissa and Garsen, two sites which are located 385km apart along the river. As no permanent tributaries are joining the main river along that stretch, it provides an ideal setup to monitor incoming and outgoing fluxes of sediment and C and to infer the processes which might be responsible for changes in the fluxes. The most important parameters that were measured, are the concentration of total suspended sediment (TSM), particulate organic carbon (POC), dissolved organic carbon (DOC) and total alkalinity and pCO2 (the partial pressure of CO2) to calculate the concentration of dissolved inorganic carbon (DIC) and the stable isotope ratios of those C pools. Additional incubation experiments were carried out during the last two campaigns whereby the changes in DOC concentration and stable isotope signature were quantified over 8 to 10 days. Besides the wet season campaigns, a daily TSM monitoring was set up at Garissa, Tana River Primate Reserve (TRPR) and Garsen between May 2012 and March 2014. The analysis of the daily sediment series revealed a strong contrast between the rating curves under non-flooded and flooded conditions. Anti-clockwise hysteresis was observed during individual discharge peaks during non-flooded conditions, while clockwise hysteresis was detected during a seasonal discharge pulse which lead to flooding, as the TSM concentrations dropped considerably at very high discharge. The calculation of present day suspended sediment fluxes in Garissa, on average 6.7 Mt yr−1 over the period 2009-2011, were of a similar order of magnitude as the estimates from before the construction of large reservoirs in the upper catchment (8.8 Mt yr−1, over various years with observations), despite an estimated sediment retention in the reservoirs of ca. 8.0 Mt yr−1. The mobilization pattern and flux stability over time was explained by autogenic processes of river bed dynamics and river bank erosion, which provided sufficient sediment to keep the river near transport capacity prior to and following dam construction. The response in the C concentration on an increase in discharge varied significantly between non-flooded and flooded conditions. The concentrations of POC and DIC at both sites were positively correlated with the discharge under non-flooded conditions. A significant amount of POC was deposited or respired upon transport, while the DIC flux systematically increased downstream. During flooded conditions, POC concentrations were low at both sites, while DIC concentrations were low upstream in Garissa and high in downstream Garsen. There was also a strong increase in DOC concentrations downstream of the flooded areas. The importance of riverine C exchange with the floodplain, the atmosphere and between the different C pools in the river was clearly visualized by closing the C budget in flux diagrams. A good estimate of POC deposition rates has been identified as a crucial factor for realistic C budgets. While the DOC fluxes were relatively small, the reactivity of the DOC is important as has been indicated in the incubation experiments: within 24-48 h, between 10 and 60% of the initial DOC pool was respired, with a simultaneous depletion in 13C of the remaining DOC. The construction of rating curves by a Bootstrap method and applying them on 73 years of discharge measurements, provides insight in the long-term fluxes under the assumption that the current C dynamics are also representative for the past or future. Under almost all annual water fluxes, deposition of sediment takes place between the two sites, while retention of C only takes place at higher annual discharge. Yet, summing the net fluxes over all the simulated years, resulted for sediment and all the C species in a overall net retention between both sites. Finally, a prototype reactive-transfer model was constructed to gain insight in the various interactions between the different biogeochemical reactions. The five tuning parameters were the deposition rate of sediment, the ratios of POC and DOC that are respired during one time step and two parameters regulating the exchange between DOC and POC. Steady-state profiles revealed the impact of each of the parameters on the longitudinal profile of the TSM and C concentrations. In a time-dependent model, two seasonal peaks of different amplitude were created as upstream input for discharge and TSM. The spatiotemporal variation in TSM and POC revealed the importance of longitudinal transport for explaining the trends particulate matter concentration, whereas the variations in DOC and DIC concentrations were clearly influenced by both the reactive components as well as longitudinal transport.Summary Samenvatting Abbreviations Contents CHAPTER 1 Introduction CHAPTER 2 Background: Study area, hydrology and methodology CHAPTER 3 Sediment yield of the lower Tana River, Kenya, is insensitive to dam construction: sediment mobilization processes in a semi-arid tropical river system CHAPTER 4 Contrasting carbon dynamics in a tropical river system (Tana River, Kenya) between flooded and non-flooded wet seasons CHAPTER 5 Dissolved organic carbon lability and stable isotope shifts during microbial decomposition in a tropical river system CHAPTER 6 Intra– and interannual variations in carbon fluxes in a tropical river system (Tana River, Kenya) CHAPTER 7 Exploring the potential of biogeochemical modeling of the carbon dynamics of a tropical river system with a reactive transfer model CHAPTER 8 Discussion, conclusion and outlook Appendix A Wet season campaigns Appendix B Incubation experiments Appendix C Discharge measurements Appendix D Model code References Curriculum vitaenrpages: 252status: publishe